Hypergravity and Categorical Feynmanology

HYPERGRAVITY AND CATEGORICAL FEYNMANOLOGY

by Louis Crane

Mathematics department, KSU.

Abstract: We propose a new line of attack to create a finite quantum
theory which includes general relativity and (possibly) the standard model in
its low energy limit. The theory would emerge naturally from the categorical
approach. The traces of morphisms from the category of representations we use
to construct the state sum also admit interpretation as Feynman diagrams, so
the non-categorically minded physicist may think of the models as combinatorial
expressions in Feynman integrals, reflecting the topology of the triangulated
manifold. The Feynman picture of the vacuum would appear as a low energy
limit of the theory.

The fundamental dynamics of the theory are determined by a Topological
Quantum Field Theory expanded around a conjectured geometric quasi-vacuum.

Although the model is a 4 dimensional state sum, it would have a stringlike
ten dimensional perturbation theory.

The motivation for the name "hypergravity" is the existence of an infinite
tower of alternating fermionic and bosonic partners of the gravitational field in
the theory, with an n=2 chiral supersymmetry functor connecting them. It is
remarkable that a supersymmetry appears in the high energy sector of a theory
not founded on supergeometry.

1. Introduction

The purpose of this paper is to begin a new attempt to produce a quantum
theory of nature, including both general relativity and particle physics, out of
the categorical approach. [CKY1,BC1,BC2].

The categorical approach is a method to produce a type of discrete theory,
rather like a lattice QFT, but put on a triangulated manifold, instead of a
lattice, in order to make it applicable to curved spacetimc.

One replaces the fields and Lagrangian of a QFT by an abstract algebraic
structure and its operations. A basis for the algebraic structure is used to label
sites on the triangulation, and the operations are used to combine these to make
numbers called "local contributions" .These are then combined into a state sum,
which can be viewed as closely analogous to a path integral. The analogy is
that the elements of the structure are like local fields, while the structure itself
is a sort of Lagrangian density.

In higher dimension than 2, the algebraic structure is often some sort of ten-
sor category and the structure which is used to create the QFT is its associator
or something similar.

The local contributions from the relevant tensor category are in fact the
evaluations of Feynman diagrams for a special QFT on hyperbolic space [F-
K,BC2]. As we shall argue below, this suggests that the Feynmanology of

1

the matter fields in the model we are proposing may appear directly as traces
of morphisms in the category of representations from which we construct the
theory.

The current paper is not self contained. It draws heavily on the categorical
state sum models for BAF theory in [CKY1], and the Euclidean and Lorentzian
signature state sum models in [BC1,BC2]. We assume that the reader is familiar
with those constructions, as well as the fundamentals of category theory. The
entire motivation for the proposal is specific features of these models, and of the
tensor category from which they are constructed.

The construction of a four dimensional topological state sum (TSS) in [CKY1]
assumes a complete tensor category, i.e. one which is closed under the tensor
product. In [BC1,2], models were proposed for four dimensional discrete quan-
tum general relativity in Euclidean and Lorentzian signatures respectively. Both
models were constrained versions of models of the topological type in [CKY1],
which were constructed by restricting to a suitable subcategory which was not
a tensor subcategory. The Euclidean model [BC1] was obtained by restricting
to the irreducible representations of U q so{4) which had equal half integer in-
dices for the left and right actions of U q so(3) , which we call "balanced" . The
Lorentzian model was a similar restriction of the category of representations of
the Quantum Lorentz Algebra [BR 1,2], specifically to the irreducible represen-
tations of the form i?o,p- (the general irreducible representation of the QLA is
of the form Rk tP for k a half integer.) Since these models are not based on closed
tensor categories, they are not topological, i.e. they depend on the triangulation
used.

In both cases the construction is motivated by the idea that the constraint
corresponds to the quantization of the condition that the bivectors on the 2-
simplices of a triangulated 4-manifold are simple.

This leaves two difficult problems in the way of developing a serious physical
theory:

1. finding the limiting behavior as the triangulation is refined, i.e.
the continuum limit

and

2. adding the matter fields.

The behavior of the topological model as the triangulation is refined is not
problematical at all. The physical quantities arc triangulation independent, so
we are already at the continuum limit. The price we pay for this is that there
are no local degrees of freedom, not even geometrical ones. The categorical
constraint has the effect of breaking topological invariance, so the "smaller"
state sum has a "larger" state space.

2

Several authors have tried to make a continuum limit for the models of
[BC1,2] or closely related ones by summing over triangulations or slightly more
general combinatorial situations [R-Rl,Ba]. The convergence problem for such
an approach seems formidable, however.

Our hope is that the two problems have a single solution.

What we want to propose is that the fundamental dynamics of our universe
is actually determined by a categorical topological state sum, but that we are
in a sort of "bubble", i.e. a region in which labels in the sum arc restricted
to a subspace of the full labelling set of objects of the category which has a
relative stability under the dynamics induced by the TSS. This is similar to
the suggestion of Witten [W] that the fundamental theory has a "topological
phase" but we fluctuated out of it, except that it is rather a TQFT that has a
nontopological quasiphase.

The constrained subtheory would be a summation over some of the labels for
the whole theory, but subject to a constraint which has the property that if we
begin a state sum with initial conditions satisfying the constraint, the dynamics
of the larger TSS would only produce a very small contribution of labels outside
the constraint for a long time as measured by a clock constructed within the
constrained label space, provided the curvature of the geometry defined by the
labelling within the constrained subsystem was small compared to the Planck
scale.

Let us formalize matters:

"DEFINITION": A geometric quasi- vacuum is a subspace S of the
total labelling space T for a TSS such that

1. The constrained labellings can be interpreted as defining a discrete ge-
ometry on the underlying manifold.

2. If we calculate the time evolution using the entire TSS, but on initial
conditions inside the subspace S, the deviation from the result of the constrained
theory will be small provided the curvature of the initial conditions is small in
Planck units, for a long time as measured by a clock defined in terms of labels
from the set S.

CONJECTURE 1: The Euclidean signature model of [BC1] is a geometric
quasi-vacuum for the TSS of [CKY1] with category Rep U q so{4) , i.e. the space
of balanced labels is a suitable S in the above definition. The Lorentzian model
in [BC2], but with q taken to a root of unity, is a geometrical quasi vacuum of
a TSS associated to a version of the QLA with q also a root of unity.

CONJECTURE 2: The set of geometric quasi-vacua associated to the
TSSs of conjecture 1 is actually very large, including examples related to string
vacua.

3

Motivation for these conjectures will appear in the bulk of the paper below.

What we are proposing is that the matter fields we see in the universe are
the low energy limit of the other labels in the TSS, besides those that satisfy
the constraints for the model for general relativity. If we are in a "bubble" , i.e.
a region where the labels satisfying the constraints predominate, small fluctu-
ations of the extra labels would appear to propagate through the geometrical
background defined by the included labels.

An observer in a region of a TSS which fluctuated into a geometric quasi
vacuum would seem to see the world obey the laws of the constrained theory.
In particular if the constrained theory approximated the laws of GR coupled to
suitable matter, it could easily settle into an expanding solution, so a Planck
scale initial fluctuation could appear to become vastly larger to an internal ob-
server. If such an observer attempted to study a region where spacetime was
highly curved, then the labels excluded from the constraint would begin to ap-
pear, adding at first to the field content of the observed theory. Paradoxically,
however, if the observer entered a region which strayed too far from the curva-
ture limitations, and hence from the constrained theory, the additional hierarchy
of "fields" would produce a regime in which no local excitations could exist at
all anymore, i.e. a topological theory. One way to understand how this would
happen is that the new sets of labels which would appear are linked by a su-
pcrsymmetry functor, so that the corrections which would appear if one refined
the triangulation would cancel out in a telescoping supersymmetry.

The way I envision this is that what would appear would be a thermal bath
of an ever increasing variety of particles, as predicted by the semiclassical theory
in regions of curvature. Eventuallythis would reach a critical temperature and
make all measurement impossible, destroying any observer in the region.

Put differently, there would not exist a true continuum quantum theory
of pure gravity, because at sufficiently high energy densities matter would be
formed around the curvature loci. Matter plus gravity would not have a contin-
uum quantum theory, because at still higher temperature the matter fields and
metric fields would merge into a topological state sum. Since the higher part-
ners of the labels which give gravity are related by a supersymmetry functor,
their contributions to the state sum would telescopically cancel as we refined
the triangulation into the region where they became important. Thus, an effect
analogous to the no renormalization theorem for supersymmetric theories would
cause the state sum to become topological.

This is rather similar to the idea of SUSY breaking, except in a discrete
picture, rather than for continuum fields. It is remarkable that the category of
unitary representations of the ordinary Lorentz group has a decomposition into
partners of the subcategory associated to GR connected by a supersymmetry
functor. Infinite supersymmetric hierarchies do not seem to appear in continuum
field theories.

The region of spacetime before the fluctuation happened would seem to have

4

a completely degenerate geometry and a thermal state at the Planck tempera-
ture to an observer inside the bubble. This would be a plausible scenario for a
big bang.

Any region inside the bubble where the curvature became high enough would
also begin to see labels from outside the constraint space, which would appear
similar to Hawking radiation, until the Planck temperature was reached.

A region within the bubble which appeared so small that any subdivision of
it would be dominated by labcllings with large curvature would effectively leave
the bubble. An attempt by an observer to probe it would yield a thermal state,
and no finer determinateness as to the geometry of the small region. Thus the
only continuum limit for the theory would be the topological theory.

The "definition" of a geometric quasi-vacuum given above is motivated by
a physical picture, and will need considerable refinement to make it mathemat-
ically precise enough for further analysis. We will try to argue that various
specific facts about the models of [CKY1 BC1,2] make this picture plausible. It
is with quasi- vacua of the specific TSS related to the unitary represen-
tations of the QLA and U q so(A) that we shall be concerned. For brevity,
and for reasons discussed below we shall refer to them as "hypergravity" .

In order to advance the program of this paper, the first necessary step will
be to make a precise formulation of the definition of quasi-vacuum in terms
of the actual categorical state sum. A natural procedure would be to con-
strain the labels along a 3D hypcrsurface in the triangulated 4-manifold to lie
within the theory, and to constrain the labels on the 2-simplices incident to the
hypersurface both to lie within the constraint subcategory and to correspond
geometrically to a small time. One could investigate the behaviour of the state
sum for some relatively simple constraint subcategory, then try to prove a quan-
titative form of the conjecture. Lastly, one could try to generalize the behaviour
to a large family of constraint subcategories. We are then conjecturing that a
very large family of subspaces of the set of labels would satisfy such a stability
condition. If our intuition is correct, we would end up with a picture of the
vacuum of the TQFT as a froth of many different types of bubble.

What we want to do in this paper is to make a preliminary analysis of the
prospects of this direction of development. There are two points which seem to
us to make it worthy of study.

1. There seems to be a natural picture in which the low energy fields pro-
duced by the excluded labels would reproduce the standard model. This is
because the algebraic structure of the category we are proposing to use makes
very natural contact with the ideas of the Connes-Chamseddinc model [CC,
Coq].

2. A perturbative treatment of the hypergravity model yields a picture of
strings with either bosonic or fermionic labels moving through a 10 dimensional

5

curved space. It is not clear yet if these are related to superstrings as we know
them. Nevertheless, this is the motivation for conjecture 2.

In the rest of this paper we will have to treat specific aspects of the CSS's
mentioned above, in order to see to what extent the conjectures are justified.

2. Representation theory of the QLA and the Hypergravity Mul-
tiplet. Constraints and Quasi- Vacua.

Let us begin by reviewing the basic facts concerning the unitary representa-
tions of the classical Lie algebra so(3,l) which is isomorphic to sl(2, C)r. These
were first studied by Gelfand, and his collaborator Naimark [GN,N]. There is a
fundamental family of these, called the principal series. They can be realized as
weighted actions of the group of linear fractional transformations on the Hilbert
space L 2 (C). They form a two parameter family Rk,p, where k is a half integer
and p is a real number. For any three representations Rki, Pi , i=l,2,3; there is
a unique intertwining operator Rkx, Pl <S> Rk 2 ,p 2 ~~ * Rk 3 ,p 3 up to scalar multiple
if ki + k 2 + k 3 is an integer, and there is no intertwiner at all otherwise. The
intertwiners are given explicitly in [N].

The constraint for the model of [BC2] is that k=0; i.e. we consider only a
state integral over the 10J symbols for the i?o, P 's.

In order to get a finite model in [BC2], we passed to representations of the
quantum group associated to the Lorentz group, the QLA of [P-Wo, Pu, BR
1,2]. The representation theory of the QLA, for a real deformation parameter q
is well analyzed in the above papers, and is quite similar to the picture for the
ordinary Lorentz algebra, except that the continuous parameter p only takes
values in the interval [0, ; ^J . The picture for intertwiners is also analogous to
the result cited for the Lorentz algebra. The result was that it was possible to
write down a state integral model using the representations of the QLA, with
all the integrals over finite measures, and hence finite.

Now we would like to think of the constrained model of [BC2] as a submodel
of a TSS. This is not quite possible for real values of q, since the sum over the
discrete parameter k would still be infinite.

However, it is natural to conjecture that if we passed to q a root of unity,
then there would be a truncation of the family of representations Rk.p at a suf-
ficiently large k for q a fixed root of unity. The reason to believe that is that
the representation Rk, P decomposes as an ascending chain of representations of
U q su{2) with the lowest spin in the chain given by k. Thus, since the represen-
tations of U q su{2) with spin greater than (lnq/4i)-l have quantum dimension
and decouple, so should the representations of the QLA at the same k. Assum-
ing this conjecture is correct, and it is certainly not too hard to investigate, We
can embed the Lorentzian model of [BC2] as a constrained subsum of a TSS.
We already know that the Euclidean signature model of [BC1] can be embedded
as a subsum of a TSS, namely the model of [CKY1] with category RepU q so{4)
for q a root of unity.

G

In either case, it is interesting to think how the rest of the labelling category
appears in relationship to the subcategory we are using as a discrete model for
gravity. We shall discuss the Lorentzian signature, the Euclidean situation is
very similar. The labelling objects appear as a series of copies of the labels for
gravity, one partner set Rk, p for each half integer k. The couplings for each
partner set would be similar to the ones in the gravity labelling set, but different
sets would be coupled more weakly to one another by some exponential factor
involving the difference in k [BR 2].

Thus we would find a hierarchy of "partners" to gravity, alternately bosonic
and fermionic.

There is also a natural operation in the category which connects the adja-
cent partner sets. Tcnsoring with a finite dimensional representation (necessarily
nonunitary) then projecting back to the unitary category would send each Ru, P
to a combination of other k's determined by the usual Clebsch Gordon formula,
but only with the identical p [BR3]. In particular, tensoring with either the
(1/2,0) or (0,1/2) representations (the complexification of the QLA is isomor-
phic to the product of 2 copies of U q sl(2, C) ) would give an operation on the
hierarchy of partner sets, mapping each Rk yP to Rk+i/2,p © Rk-i/2, P - Since ten-
sor product is functorial, the morphisms of the category, which we are using to
construct our diagrams, are also acted on by the supersymmetry functors.

This is the motivation for the name "hypergravity" . If we think of the
labelling category for the TSS as an extension of the labels for the model for
GR, it has the appearance of a hierarchy of copies of the GR labels, alternatively
fermionic and bosonic, together with two natural chiral fermionic maps relating
adjacent elements of the hierarchy. This is reminiscent of supersymmetric field
theories, except that the hierarchy seems to be infinite. The coupling of each
"hyperpartner" to itself mirrors the gravity multiplet.

The above cited result about coupling of different partner sets is one mo-
tivation for conjecture 1 above. For small curvatures, it would be improbable
to see k fluctuate by an entire half unit. Another piece of evidence is the fact
that if we restrict the models of [BC1,2] to configurations corresponding to flat
metrics, they in fact become topological themselves [CKY1]. Of course the idea
that "nearly flat" metrics are "nearly topological" needs a careful quantitative
study.

At this point, I hope it is clear that "what would be the low energy effective
field content of the hypergravity multiplet?" is a plausible question. One could
try to find an analog of the renormalization group for categorical state sums,
and try to discover how much of the hypergravity multiplet was relevant.

In the next section we explore the possibility that, because of an algebraic
coincidence of noncommutative geometry which does not appear to have a classi-
cal geometric analogue, the standard model might well emerge naturally within
hypergravity.

3. Connections with the Standard Model and the Connes-Lott

7

Model

In [C-C] , it was argued that the standard model emerges naturally provided
an algebra which the authors called the "world algebra" appeared as a symmetry
algebra. The algebra they meant was

W = C®H®M 3 (C) .

It was incidentally mentioned in their work, and later elaborated in [Coq],
that this algebra emerges as the semisimple part of the quantum group U q sl(2, C)
when we let q be a third root of unity.

Note that this is different from the common truncation at a 4rth root of
unity. This will be still another new algebraic question to study in order to
investigate this model.

Now let us remind ourselves that the QLA is a "real form" of U q sl(2, C) <S>
U q sl(2,C), more precisely, that its complexification is exactly that algebra
[BR1].

Hence a truncation of the QLA at a third root of unity would contain two
chiral copies of the "world algebra" as its semisimple part.

It is not unreasonable to expect that a nilpotent piece of an algebra would
vanish in a low energy (=long distance) limit, leaving the world algebra as the
effective symmetry of the theory.

This suggests that it might be interesting to examine other odd roots of
unity as well, to see if they predict hierarchies of fields which include both
the standard model and others which might be more massive, but potentially
discoverable.

It is not possible at this point to demonstrate that the standard model
actually emerges from the hypergravity multiplet at q a third root of unity.
The truncated representation theory needs to be studied, to determine how
the representations of our category decompose under the action of the world
algebra. In addition, a form of the renormalization group needs to be developed
for CSS's, and many hard calculations need to be done. Nevertheless, in QFT
symmetry is a very strong principle, and seeing the symmetry of the standard
model emerge by accident in the hypergravity picture is very surprizing at least.

4. Categorical Feynmanology

In our proposal, there is no true (continuum) QFT except TQFT. The la-
bels that appear as matter fields in a certain regime merge into a topological
sum in the continuum limit. Thus the effective theory of matter should be a
combinatorial, algebraic one, which could appear within a tensor category.

Such a point of view already exists in QFT as practiced, namely, that of
Feynman.

8

It has long been one of the disturbing problems of theoretical QFT that
neither the derivation of the Feynman rules from a QFT nor the reconstruction
of a QFT from Feynman amplitudes is mathematically sound. On the other
hand, it is rather the Feynman rules than the continuum QFT which have
directly been verified by experiment.

All this makes us wonder if the Feynman diagrams of the physically relevant
QFTs could appear directly within the tensor categories we are using in our
constructions.

This proposal is not out of harmony with the original ideas of Feynman
[F], who seems to have believed that fundamental quantum processes were dis-
crete, and to have been rather sceptical of continuum QFT as a framework.
The point of view that perturbativc QFT simply sidestepped unknown physics
that appears at a very high energy has been a popular one for the subsequent
development of perturbation theory.

We are proposing that a categorical state sum is the most natural candidate
for the unknown high energy Physics.

We want to present some lines of argument to make this plausible.

In the first place, the Feynman rules have a very close formal similarity
to the structure of a tensor category. Morphisms in a tensor category can
be constructed using graphs, with objects on edges and tensor operators on
vertices. Furthermore, duality in tensor categories gives an identification of the
set of vertices with a certain number of lines in and the rest out with the vertices
with the same number and type of lines, but the separation into in and out lines
changed. For example, Hom(A ® B,C ® D) = Hom(A <g> C* , B* ® D). This is
formally identical to Feynman's use of CPT invariance to identify vertices with
edges going forward in time with similar edges going backwards.

The similarity between Feynmanology and categorical diagrams is not ac-
cidental. When one constructs a Lagrangian, one first determines what fields
one needs to include, then arranges them as representations of the symmetries
the theory is expected to have. Next, one looks for scalar combinations of the
fields to include in the Lagrangian. These provide the possible vertices of the
Feynmanology of the theory.

Note that representations of a group or algebra form a tensor category, and
that by duality, scalars in the theory correspond to tensor operators. Thus,
up to this point, the construction of tensor categories and Feynmanologies is
identical.

Then there is the question of the Feynman integral, which gives us an evalua-
tion of a Feynman diagram. In [B, F-K], it was pointed out that the evaluations
of the closed diagrams in the categories of representations for Euclidean 4D GR
(relativistic spin nets), were exactly Feynman integrals. In [BC2, R-Rl ], the
same was shown for a regularized definition of an evaluation for closed diagrams
in the category of unitary representations of the Lorcntz algebra.

It might seem that the analogy between Feynmanology and tensor categories
breaks down when we evaluate an open Feynman graph and get a number rather

9

than a morphism. This is not correct. The evaluation of a Feynman graph is
a function of momenta for the incoming and outgoing particles. Passing to the
momentum representation for the state spaces, we can regard these numbers
as the matrix elements of a linear map. It is natural to think of this map as
a morphism in a tensor category. The Ward or Slavnov- Taylor relations [t'H]
amount to the statement that this map is an intertwiner for the gauge algebra,
i.e. that it lives in the category of representations of the total symmetry of the
system.

In the standard construction of a QFT, we would now select a few fields as
the fundamental ones, find which vertices are renormalisable, and adjust the
coupling constants in front of them to fit experiment. The picture of the vac-
uum in Feynmanological QFT is then in effect a subsummation of a categorical
state sum, in which only special terms are counted, and with special numerical
weights.

We now want to make a specific conjecture as to how Feynmanology as we
know it arises.

CONJECTURE OF CATEGORICAL FEYNMANOLOGY: Let us
postulate that just below the scale at which the hypergravity model
goes topological, we have all the objects and morphisms of the cat-
egory of representations of the physical symmetries of the standard
model appearing in an effective state sum. This would correspond
to a sum over Feynman diagrams, but allowing all possible particles
and vertices. Let us then assume that the renormalization group acts
on this theory, making all the unobserved types of particles become
unstable in the low energy limit, and all the unrenormalizable ver-
tices irrelevant (i.e. shrinking their coupling constants to near 0) and
giving the remaining coupling constants their physical values.

The Feynman picture of the vacuum then appears as the low en-
ergy surviving remnant of the full categorical sum.

The remarkable thing about this conjecture is that it is really not so terribly
radical. The idea that "all terms not forbidden by a symmetry appear in the
effective Lagrangian" is commonplace in QFT. Particles are considered to be
completely specified by their "quantum numbers" which just specify the repre-
sentation of the physical symmetry which they form. The idea that the observed
interactions are determined by the renormalization group is also standard.

It is not clear whether the conjecture above, which boils dow to the assump-
tion that all possible fields appear at the fundamental scale and all vertices
appear with equal coupling constants, has any physically testable consequences.
It is arguable that it removes an esthetic problem from QFT. There has been
the feeling that the Feynman revolution of the late 40s was disturbing in that
it consisted of calculational tricks rather than a conceptual leap. If we really
should think of summing over the entire category of representations of the total
physical symmetry rather than merely the low energy limit, then Feynman in
effect discovered the relationship between combinatorial topology and categor-

10

ical algebra, which in some eyes is an elegant and fundamental departure from
continuum Physics after all.

We are led to the idea that the Feynmanology of the standard model could
appear directly as terms in our hypergravity TSS corresponding to represen-
tations of the world algebra, which, if our conjectures are true, would become
important in an expansion around the constrained model in a low energy regime.

This suggests that although the strong force is nonperturbative in the in-
frared, physical calculations using perturbation theory for QCD could actually
be done, since the onset of the topological theory would cut off the perturba-
tion series. Terms with a large but finite number of loops would dominate the
physics. Summing them would be equivalent to a categorical state sum on a
triangulation with Planck scale spacings.

It is very tempting, in such a model, to take the ideas of Feynman more lit-
erally. Perhaps we might think of modifying our quasi-vacuum to include terms
which fill the Dirac sea. It is also very natural to think of inserting Feynman
vertices, with their invariance wrt time ordering into state sums representing
general relativity, where different terms would correspond to different causal
orderings.

This picture suggests that we attempt to generalize Feynmanology to the
categorical setting. We could attempt to investigate such fundamental ideas
of particle physics as renormalizability, anomalies, and current algebras in the
greater generalization of a tensor category for example, using representations of
a quantum group. Such a study would be inherently interesting, and might also
help us to further the program of this paper.

5. Stringy Vacua

It was observed by Lee Smolin [S], that the natural perturbation theory
around a classical solution in the Euclidean signature categorical state sums we
are studying would resemble propagation of a string. This is because the natural
basic perturbation to make to a labelled diagram representing a morphism in
the category of representations of U q su(2) is to tensor by a constant spin around
a loop. In fact, the binor calculus shows that any "spin net" as such diagrams
are also known, is a superposition of such minimal perturbations.

It is not immediately obvious that this stringy picture would work as well
in the Lorentzian signature, but that is a common enough problem in string
theory.

Thus, if we want to study perturbation theory around a fixed term in a cate-
gorical state sum, we insert loops with spin labels, and then ask the probability
that some other combination of basic perturbation loops appears later. This
can be calculated by tracing the different possible paths for the loops, i.e. by
summing over discrete world sheets for a discretized string.

Assuming for the moment that such a perturbation theory has some range of
validity (a non-trivial assumption, but at least one which could be investigated),

11

in what sort of space would it appear?

The CSSs we are using in this paper all begin by putting representations on
the 2-simplices of a triangulated 4-manifold. Thus, the loops of representations
we use as perturbations are occupying little circles of 2-simplices. If we think
of the simplices as small, we can approximate these as circles in the bundle of
bivectors over the spacetime 4-manifold.

Now the bundle of bivectors over a 4-manifold is a 10 dimensional manifold,
fibered over the spacetime with a 6 dimensional fiber.

The natural suggestion is that in the limit of finer triangulations the per-
turbation theory would look like 10 dimensional string theory. Since the hy-
pcrgravity multiplet contains both bosonic and fermionic terms with a natural
functor connecting them, it is tempting to think it would reproduce superstring
theory.

This further suggests that geometries in 10 dimensions could serve as qua-
sivacua for the theory. One would decide what representation to assign to a
2-simplex by finding its area viewed as a surface in the bivector space endowed
with a suitable metric.

We seem to be suggesting that the extra 6 dimensions for compactified mod-
els should be thought of not merely as small, but as infinitesimal, i.e. as bivectors
associated to the tangent space.

This conjecture is principally motivated by the enormous work already ex-
isting on superstrings and string vacua. The only connections that are really
solid at this point are the coincidence of dimensions, the hierarchy of bosonic
and fermionic excitations, and the "stringiness" of the picture.

6 Generalizations. Membrane theory?

In broad outline, we are suggesting that the combination of a TSS and
some interesting constrained subtheory might provide a mathematical setting
for quantum physics. We have, in effect, begun to study the simplest possible
such combination, corresponding to the lowest rank of noncompact quantum
groups. It is intriguing that there is some indication that the standard model
might emerge from the simplest example.

However, there will certainly be many other similar models which one could
investigate. If the idea of obtaining the world algebra from the hypergravity
multiplet didnt work, one could try bringing a gauge group directly into the
theory by starting with a larger noncompact quantum group, for example. If
nothing else, this would pose some very natural questions for representation
theory.

Another possible generalization comes from the fact that the four dimen-
sional TSS of [CKY1] from which we started is not the most general categorical
construction possible. This is because it begins from a braided tensor cate-
gory, whereas the most general 4D construction would use a spherical tensor
2-category [Ml, 2] or else a Hopf category [CF]. The construction we are using

12

is a special case, because a braided tensor category is a 2-category with one
object.

The most general construction puts objects from the 2-category on edges of
the triangulation, and 1-morphisms on faces. Since a braided tensor category
is a 2-category with one object, the labelling of the edges becomes trivial and
disappears.

In terms of the diagrammatic description, the edges of the diagram become
boundaries of surfaces, which carry new labels, in the most general case.

At this point, I do not know of any physically significant tensor 2-categories.
Interesting mathematical examples are constructed in [M2].

However, it is amusing to ask what would happen if we tried to find a
geometric picture for a perturbation theory for a TSS model based on a general
spherical tensor 2-category. The edges would correspond to bivectors as above,
while the surfaces bounding them would live in the space of trivectors containing
a bivector. The total space of such configurations is a 12 dimensional space.
Thus, something like a membrane theory in 12d would be the natural choice for
a geometric background. This is of course extremely speculative, but perhaps
worth mentioning.

7. Conclusions. A Convergence of Ideas?

The direction we are mapping out is based on four ideas:

1. The fundamental geometry of spacctimc is discrete, not continuous, hence
should be constructed from a model on a triangulated manifold.

2. The model should be constructed from the structure of a tensor category,
or if you prefer from Feynman integrals attached to the simplices.

3. The continuum limit should be fixed by the coming into play of a hier-
archy of new fields, corresponding to more parts of the tensor category, which
remove the ultraviolet problems by including an entire tensor category and hence
becoming topological.

4. The largeness, flatness, and apparent complexity of the universe as we
see it is due to an initial fluctuation, which caused it to take an apparent form
more complex than the fundamental one.

Aside from the categorical formulation, these ideas are not new in theoretical
physics. In particular, the picture of a multiplet of fields which comes into play
only at high energies and softens out short distance behavior is very similar to
supersymmetric theories.

What is new here is the application of unitary representations of noncompact
groups, and the proposal to use the structure of the tensor category they form,
and q-deformations of it directly to generate a physical model.

13

The model we are proposing is quite close to the spin foam picture [R-
R2, Ba], which itself grew out of the loop variable program [R-S]. The most
important difference between the two is the passage from a three dimensional
to a four dimensional symmetry group, due to the adoption of a Lagrangian
approach.

To the (perhaps somewhat partial) eye of the author, one of the remarkable
features of the hypergravity model is the number of earlier ideas of twentieth
century Physics to which it seems to give a home.

In addition to the spin foam/loop variables picture mentioned above, and to
the possible connection with superstring theory, one could mention the old idea
of Einstein [St] that the fundamental theory should be discrete and algebraic,
Feynman's recasting of particle Physics as a discrete theory (as we can say now,
of evaluations of words in a tensor category), Witten's idea of a "topological
phase" , and Connes derivation of the standard model.

Another intriguing connection is with the ideas of Majid [Maj]. Since the
QLA is the quantum double of a quasitriangular Hopf algebra, it is a very
canonical example of the crossed bialgebras he studies. Its role in the models
in this paper is therefore reminiscent of the suggestions he makes relating them
to quantum gravity. I do not know if his deeper ideas about duality will appear
more directly in the development of this program.

Perhaps, even though each of these connections separately needs considerable
work to make it more precise, the sheer number of natural connections to other
streams of thought is worth noting.

We have made a large conceptual shift here from the mainstream of quantum
Physics. We are replacing continuum geometry and fields by combinatorial ge-
ometry and categories of representations. The justification for any fundamental
reformulation must be that it opens new possibilities for investigation.

It is much too soon to say that this program will solve the problems of the-
oretical physics. It is fair to say that it contains many directions to explore
without obvious analogs in the continuum picture, and with suggestive connec-
tions to physical phenomenology. At a minimum, it can hardly fail to suggest
interesting new questions in categorical algebra.

BIBLIOGRAPHY

[B] J. W. Barrett, The Classical Evaluation of Relativistic Spin Networks,
Adv. Theor. Math. Phys. 2 (1998), 593-600.

[BC1] J. W. Barrett and L. Crane, J. Math. Phys. 39 (1998) 3296-3302.

[BC2] J. W. Barrett and L. Crane, A Lorentzian Signature Model for Quan-

tum General Relativity gr-qc 9904025, to appear CQG

[Ba] J. C. Baez, Spin Foam Models, Class. Quant. Grav. 15 (1998) ,
1827-1858.

14

[B-Rl] E. Buffcnoir and Ph. Roche, Harmonic Analysis on the Quantum
Lorentz group, q-alg 9710022

[B-R2] E. Buffenoir and Ph. Roche, Tensor products of Principal Unitary
Representations of the Quantum Lorentz Group and askey Wilson polynomials
math.QA 9910147|

[CF] L. Crane and I. Frenkel, Four Dimensional Topological Quantum Field
theory, Hopf Categories and the canonical Bases, J. Math. Phys. 35 (1994),
5136-5154.

[C-C] A. Connes and A. Chamseddine The Spectral Action Principle hepth
9606001

[Coq] R. Coquereaux, On The Finite Dimensional Quantum Group M3 +
M 2ll {Lambda 2 ) Lett. Math. Phys. 42 (1997) 309-328

[CKY1] L. Crane, L. Kauffman, and D. Yetter, State Sum Invariants of 4-
Manifolds, JKTR 6(2) (1997), 177-234

[CY2] L. Crane and D. Yetter, unpublished.

[F] R. Feynman, Theory of Fundamental Processes W. A. Benjamin, 1962

[F-K] L. Frcidcl and K. Krasnov, Simple Spin Networks as Feynman Graphs.
tiep-th/9903192

[G-N] I.M. Gelfand and M. A. Naimark, Unitary representations of the
proper lorentz group, Izv. Akad. Nauk. SSSR 11 411 (1947)

[Ml] M. Mackaay, Spherical 2-categories and 4-Manifold Invariants, Adv.
Math. 143 (1999), 288-348.

[M2] M . Mackaay, Finite Groups, Spherical 2-Cate gories and 4-Manifold
Invariants, math-qa 9903003, to appear Adv. Math.

[Maj] S. Majid, Foundations of Quantum Group Theory Cambridge Univer-
sity Press 1995

[N] M. A. Naimark, Decomposition of a tensor product of irreducible rep-
resentations of the proper lorentz group into irreducible representations, Am.
Math. Soc. Translations ser 2 v 36

[P]W. Pusz, Irreducible unitary representations of the quantum Lorentz
group, Comm. Math. Phys. 152 571-626 (1993)

15

[P-Wo] P. Podles and S. L. Woronowitz Quantum deformations of the Lorentz
Group Comm. math, phys 136 399-432, (1990)

[R-Rl] M. P. Reisenberger and C. Rovelli, Sum Over Surfaces Form of Loop
Quantum Gravity, Phys Rev. D56 (1997) 3490-3508

[R-R2] M. P. Reisenberger and C. Rovelli, Sum over Surfaces Form of Loop
Quantum Gravity, Phys. Rev. D 56 (1997), 3490-3508.

[R-S] C, Rovelli and L. Smolin, Spin Networks and Quantum Gravity Phys.
Rev. D 52 (1995) 5743-5759

[S] L. Smolin, Strings as Perturbations of Evolving Spin Networks hepth
9801022

[St] J. Stachel, Einstein and Quantum Mechanics in Conceptual Problems
in Quantum Gravity Birkhauser 1991

[t'H] G. t'Hooft and M. Veltman diagrammar in Under the spell of The gauge
Principle, World scientific, 1994

[W] E. WittenTopological Quantum Field Theory and The Jones Polyno-
mial, Comm. Math. Phys. 121 (1989) 357

16